Machine Learning Applications in Search Algorithms for Gravitational Waves from Compact Binary Mergers - Dissertation Abstract

Aug 4, 2026

Upload by Jiven

Gravitational waves from compact binary mergers are routinely observed by Earth-based detectors, enabling new insights into the Universe. Extracting signals from noisy data relies on matched filtering with template banks, but increasing detector sensitivity and richer waveform models expand the number of templates and can become computationally prohibitive, especially for low-latency alerts. This thesis evaluates deep learning search algorithms, compares them using sensitive distance, and introduces methods for binary black hole and binary neutron star detections at previously untested statistical confidence levels.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages287

Reading Time12 h 3 min

AccessFree to download

Abstract

Contents

Introduction

Foundations

Deep Learning

Detection of Gravitational-Wave Signals from Binary Neutron Star Mergers

Gravitational-wave Merger Forecasting

Training Strategies for Deep Learning Gravitational-Wave Searches

From One to Many: A Deep Learning Coincident Gravitational-Wave Search

Machine Learning Applications in Search Algorithms for Gravitational Waves from Compact Binary Mergers Von der QUEST-Leibniz-Forschungsschule der Gottfried Wilhelm Leibniz Universit􀁿at Hannover zur Erlangung des akademischen Grades Doktor der Naturwissenschaften Dr. rer. nat. genehmigte Dissertation von M.Sc. Marlin Benedikt Sch􀁿afer Referent: Prof. Bruce Allen Korreferenten: Prof. Bernd Br􀁿ugmann Dr. Francesco Salemi Promotionskommission: Prof. Domenico Giulini Prof. Bruce Allen Prof. Bernd Br􀁿ugmann Tag der Promotion: 16.12.2022 Abstract Gravitational waves from compact binary mergers are now routinely observed by Earth-bound detectors. These observations enable exciting new science, as they have opened a new window to the Universe. However, extracting gravitational-wave signals from the noisy detector data is a challenging problem. The most sensitive search algorithms for compact binary mergers use matched 􀀌ltering, an algorithm that compares the data with a set of expected template signals. As detectors are upgraded and more sophisticated signal models become available, the number of required templates will increase, which can make some sources computationally prohibitive to search for. The computational cost is of particular concern when low-latency alerts should be issued to maximize the time for electromagnetic follow-up observations. One potential solution to reduce computational requirements that has started tobe explored in the last decade is machine learning. However, di􀀋erent proposed deep learning searches target varying parameter spaces and use metrics that are not always comparable to existing literature. Consequently, a clear picture of the capabilities of machine learning searches has been sorely missing. In this thesis, we closely examine the sensitivity of various deep learning gravitational-wave search algorithms and introduce new methods to detect signals from binary black hole and binary neutron star mergers at previously untested statistical con􀀌dence levels. By using the sensitive distance as our core metric, we allow for a direct comparison of our algorithms to state-ofthe-art search pipelines. As part of this thesis, we organized a global mock data challenge to create a benchmark for machine learning search algorithms targeting compact binaries. This way, the tools developed in this thesis are made available to the greater community by publishing them as open source software. Our studies show that, depending on the parameter space, deep learning gravitational-wave search algorithms are already competitive with current production search pipelines. We also 􀀌nd that strategies developed for traditional searches can be e􀀋ectively adapted to their machine learning counterparts. In regions where matched 􀀌ltering becomes computationally expensive, available deep learning algorithms are also limited in their capability. We 􀀌nd reduced sensitivity to long duration signals compared to the excellent results for short-duration binary black hole signals. Keywords: gravitational waves, compact binary mergers, deep learning, gravitational-wave search algorithms Contents Abstract iii Contents viii List of Figures x List of Tables xi List of Acronyms xiii 1 Introduction 1 2 Chapter Descriptions and Authorship Clari􀀌cations 5 3 Foundations 9 3.1 Gravitational Waves ....................... 10 3.1.1 Linearized Gravity .................... 12 3.1.2 Post-Newtonian Formalism ................ 19 3.1.3 Waveform Models ..................... 23 3.2 Data Analysis for Compact Binary Coalescence Signals .... 26 3.2.1 Noisy Detector Data ................... 28 3.2.2 Matched Filtering ..................... 37 3.2.3 Search Algorithm and Signi􀀌cance of Detections .... 43 3.3 Deep Learning ........................... 46 3.3.1 Neural Networks ..................... 48 3.3.2 Training Neural Networks ................ 54 3.3.3 Convolutional Neural Networks .............

FAQs

Why do matched-filtering template banks become computationally challenging for compact-binary searches?

As detectors are upgraded and waveform models improve, the required number of templates grows, increasing computational cost and potentially making some searches impractical for timely alerts.

How does the thesis compare different deep learning gravitational-wave search algorithms?

It uses sensitive distance as the core metric, enabling direct comparison of algorithms to state-of-the-art search pipelines.

What does the thesis contribute beyond evaluating sensitivity?

It introduces new detection methods reaching previously untested statistical confidence levels and organizes a global mock data challenge, releasing developed tools as open-source software.

Aug 4, 2026

Upload by Jiven